Key Research Interests and Methodologies:

Our work combines cutting-edge technologies, advanced computational methods, and robust infrastructure to address key challenges in neuropathology:

• Structured Data Integration: The AI-supported CNS (Coded Network Semantics) system transforms traditionally unstructured diagnostic reports into structured data formats, enabling the seamless integration of high-dimensional omics datasets, including next-generation sequencing (NGS) and DNA methylation profiles.
• AI-Driven Histological Analysis: We utilize pre-trained pathology foundation models, such as convolutional neural networks (CNNs) and attention-based architectures, to automate the analysis of histological images, predicting molecular and clinical features directly from routine hematoxylin-eosin (HE) slides.
• Explainable AI (XAI): Transparency-focused algorithms are integrated into AI workflows, ensuring that predictions are interpretable and actionable in clinical practice, fostering trust and adoption among clinicians.
• Cross-Site Data Normalization: Advanced computational techniques, such as latent space disentanglement and domain adaptation algorithms, harmonize morphomolecular data across institutions, minimizing technical variability and enabling large-scale, interoperable analyses.
• Federated Learning Neurooncology and Rare Disease: Secure federated learning frameworks allow the development of robust AI models trained on distributed datasets, overcoming the limitations of small, localized datasets for brain tumors and rare diseases like neuromuscular disorders.
• High-Performance Infrastructure: Dedicated CPU and GPU clusters support high-throughput data processing, large-scale AI model training, and federated learning, enabling the efficient analysis of omics and imaging data.